CODINGTON, DOERR,AND FOX
564
waa taken to dryness an vacuo. To the residue was added a solution of 15 g. of sodium acetate in 200 ml. of water. After stirring for a few minutes a colorless amorphous solid was collected and washed with water. The yield wm nearly quantitative. The solid waa purified by precipitating from an ether solution with petroleum ether, m.p. 9&140° (efferv.), [aIz4D+5’ ( c 0.2,acetone). Ultraviolet absorption properties in ethanol were hmax 255, 229.5 mfi; Xmln 246.5 m r ; ratio 260/230 mr 0.63. Anal. Calcd. for ClTHI7BrN2OsS:C, 41.73; H, 3.50; Br, 16.33; N, 5.72; S, 6.55. Found: C, 41.43; H, 3.67; Br, 17.05; N , 6.25; S, 7.49. 5‘-O-Benzoyl-2 ‘-chloro-Z‘-deoxy-3’-O-mesyluridine [X(Cl)] .A suspension of 1.0 g. (2.45 mmoles) of VI11 in 50 ml. of anhydrous dioxane ww saturated with dry hydrogen chloride at 0’. The resulting clear solution was allowed to stand at 20-25’ for 7 days. Solvent wm removed in vacuo, and the resulting gum was dissolved in chloroform and extracted several times with water.
Nucleosides. XIX.
VOL.29
The chloroform layer waa dried over sodium sulfate, then taken to ‘dryness in vacuo. Trituration in ether gave a colorless solid. This waa filtered, washed with ether-petroleum ether, and dried. The yield of amorphous solid, m.p. 90-130° (efferv.), [a]*‘D +3’ ( C 0.3, acetone), wm 1.07 g. (98%). Ultraviolet absorption properties in ethanol were max. a t 255 and 229 mp, emex 9180 and 12,840, respectively; min. at 247 mp, emin 5260. Anal. Calcd. for CI7Hl7C1N2O8S:C, 45.90; H, 3.86; C1, 7,98; N,6.31. Found: c,45.88; H,3.88; C1,8.01;N,6.52.
Acknowledgment.-The authors wish to thank Mrs. Dina Van Praag for valuable contributions to this work. They would like to express their appreciation to Dr. Aaron Bendich and to Dr. George B. Brown for helpful discussions and continued interest.
Structure of the 2‘-Halogeno-2’-Deoxypyrimidine Nucleosides’
JOHN F. CODINGTON, IRISL. DOERR, AND JACK J. Box The Division of Nucleoprotein Chemistry, Xloan-Kettering Institute j o T Cancer Research, Sloan-Kettering Division of Cornell University Medical College, New York 61,New York Received August 26, 1963 The halogeno deoxy nucleosides prepared by the reaction of hydrogen halides with 2,2’-anhydro pyrimidine nucleosides2 ( I ) were proven t o be 1-( 2’-halogeno-2’-deoxy-P-o-r~bofuranosyl)pyrimidines (11). 2’-Fluoro-2’deoxyuridine [II(F)] was converted to the 3’,5’-di-O-trityl derivative (XIII). Reflux of XI11 with base formed 1-(3’,5‘-di-O-trityl-p-~-arabinofuranosy~)uracil (XV). XV also was formed from the tritylation of 1-(5’-0trityl-P-D-arabinofuranosyl)uracil(XVIII). Detritylation of xv with acid gave only l-P-D-arabinofuranosyluracil (XVI), proving both the 2’-position and ribo configuration of the fluoro function in II(F). Proof of analogous structures for 2’-chloro- [II(Cl)l and 2‘-bromo- [II(Br)l 2‘-deoxyuridines was obtained by their conversion to 3’,5’-di-O-mesyl-2’-halogenoderivatives [VIIIk(Cl) and VIIIk(Br)], identical with products obtained upon reaction of the corresponding hydrogen halides with the known 2,2’-anhydro-l-(3’,5’-di-O-mesylp-D-arabinosyl)urad( IXk).
The previous paper2 in this series describes the reacwas present, even in small yield, could not be excluded, since in no case were quantitative yields of halogeno tion of 2,2‘-anhydro pyrimidine nucleosides with andeoxy nucleosides obtained. hydrous hydrogen halides to yield fluoro-, chloro-, and Although I11 was not involved in aqueous acid cleavbromodeoxy nucleosides. Although these reactions age of I (only 1-8-D-arabinofuranosyluracil was obwere expected to take the pathway illustrated by the reaction of 2,2’-anhydro-l-~-~-arabinofuranosyluraciltained as a product),’ a reaction route similar to that considered above for anhydrous acid (hydrogen halide) (I) to give 2’-halogeno-2‘-deoxy derivatives of the ribo cleavage of I (namely, I to I11 to V) was almost cerconfiguration I1 (Scheme I), another plausible pathway tainly involved when anhydrous alkaline reagents were existed. I t appeared conceivable that, under the reacemployed.8-I2 tion conditions, a 2’,3’-ribo epoxide (111) could have The present paper presents proof of structure for formed from I. Such a reaction route was supported by data on the cleavage of 2,3’-anhydro-l-/3-~-xylo- the halogeno deoxynucleosides described in the preceding paper.2 Although this investigation has infuranosyluracil (IV) in refluxing aqueous acid, to yield volved only deoxyuridine analogs, on the basis of chemi(by paper electrophoretic examination) spots correcal, physical2 and bi~logical’~ properties, it is highly sponding to 1-p-D-xylofuranosyluracil and 1-p-D-araprobable that derivatives of thymidine and 5-fluorobinofuranosyluracil.3 Reaction of hydrogen halides deoxyuridine have analogous structures. with epoxide I11 would be expected to give mixtures Despite the fact that the halogeno deoxy nucleosides of the 3’-halogeno derivative (xylo configuration, V) could be converted back to the 2,2’-anhydro starting and the 2’-halogenonucleoside (arabino configuration, VI) with the 3’-halogeno derivatives (V) predominat(7) D. M. Brown, A . R. Todd, and S. Varadarajan, J . Chem. SOC.,2388 ing.4-6 In no case were mixtures of halogeno deoxy (1956). nucleosides isolated, but the possibility that an isomer ( 8 ) Brown, et O Z . , ~ treated compound I with sodium ethyl sulfide in di(1) This investigation was supported in part by funds from the National Cancer Institute, National Institutes of Health, U. S. Public Health Service (Grant No. CA 03190-07). (2) J . F. Codington, I . L. Doerr, and J. J. Fox, J. Org. Chem., 89, 558 (1964). (3) N.Yung and J. J. Fox, J . A m . Chem. Soc., 83, 3060 (1961). (4) See C. D. .4nderson, L. Goodman, and B. R. Baker, ibid., 81, 898, (1959),and additional references therein. ( 5 ) J. F. Codington, R. Fecher, and J. J. Fox, J . Orp. Chem., 87, 163 (1962). (6) G. Casini and L. Goodman. J . A m . Chem. Soc., 86, 235 (1963).
methylformamide and obtained 1-(3’-ethylthi~-3’-deoxy-~-~-xylof uranosy1)uracil (V, (X) = -SEt).lO (9) D. M. Brown, D. B. Parihar, 4. R. Todd, and S. Varadarajan, J . Chem. SOC.,3028 (1958). (10) The transient existence of 111 under alkaline conditions has been postulated on other occasions.”~1~ (11) J. F. Codington, R. Fecher, and J. J. Fox, J . Am. Chem. SOL.,82, 2794 (1960). (12) E. J. Reiat. J . H . Osiecki, L. Goodman, and B. R. Baker, ibid., 8 8 , 2208 (1961). (13) J. F. Codington, I. L. Doerr, L. Kaplan, and J. J. Fox, Federation Proc., 88, 532 (1963).
2‘-HALOGEN0-2’-DEOXYPYRIMIDINENUCLEOSIDES
MARCH,1964
565
SCHEME I 0
0
Wck$-
OH H
m
AN
O
m
P
XI
SCHEME I1 materials (I) in each case, such conversions do not constitute absolute proof of structure. Anhydro nucleoside formation could conceivably proceed by a reversal of the route postulated for the formation of PN structures V and VI, namely, V or VI to 111, and I11 to I.14 In order to establish the position of the substituent on, and the configuration of, the halogeno deoxy nucleosides, it would be necessary to effect 2,2‘OH (XI MsO (XI MsO anhydro formation with both free hydroxyls (3’ and 5 ’ ) blocked with appropriate groups in order to rule II Ix out the possibility of intermediary epoxide formation during the reaction. In the case of 5‘-0-benzoyl-3’-0-mesyl-2’-chloro- (XI= F, CI, Br 2’-deoxyuridine [VIIIj(Cl) ] and its 2’-bromo analog VIIIj(Br) (see Scheme 11) described in the previous paper12structures were readily established, as both the 3’- and 5’-positions were blocked, and the blocking groups remained in position during anhydro formation. This was accomplished by heating VIIIj(C1 or Br) The structures of the chloro [II(Cl)] and bromo with sodium benzoate in acetamide”,l7 to give the [I1(Br) ] deoxynucleosides (Scheme I) were established known 2,2’-anhydro-1- (5‘-O-benzoyl-3‘-O-mesylarabin- without undue complication (and by analogy for the osy1)uracil (IXj).ll corresponding thymidine analogs). It was found that The establishment of a 2’-ribo position for the halothe reaction of the structurally established 2,2’-angeno substituents in VIIIj(C1 or Br) supported such a hydro-1- (3’,5’-di-0-mesyl-p-D-arabinosyl)uracil (IXk , 2‘-ribo position for the halogen atoms in the unblocked Scheme 11) with hydrogen chloride in dioxane halogeno deoxy nucleosides. It was clear, however, gave a crystalline monochlorodimesyloxy derivative that no alternate route (epoxide intermediate) could [VIIIk(Cl)]. Compound VIIIk(C1) was converted be postulated for the introduction of a chloro or bromo back to the dimesyloxy anhydro nucleoside (IXk) by substituent into IXj as was true of I, which contained means of sodium acetate in dimethylformamide, l 9 thus an unblocked 3’-hydroxyl. establishing the position and configuration of the chloro substituent in VIIIk(C1) as 2’-ribo. Mesylation of (14) A similar situation pertains t o t h e structure of a n iodo deoxy nucleoII(C1) in pyridine yielded a crystalline product idenside prepared by Brown and co-workers*6 who assigned a 5’-0-acetyl-2’tical with VIIIk(C1). Therefore, the position and iodo-2’-deoxyuridine structure. This compound was converted t o t h e 2 2 ’ anhydro derivative,Is which, upon reduction followed by deacetylation,ls configuration of the chlorine in II(C1) is 2’-ribo. gave 2’-deoxyuridine (V11). Although these investigators assigned a 2‘The product of the reaction of the 2,2‘-anhydro rib0 structure, the possibility t h a t the iodo deoxy nucleoside had a n arobino dimesyloxy nucleoside (IXk) with hydrogen bromide configuration ( V l , 5’-0-acetyl, (X) = iodo) could not be discounted. T h e formation of the epoxide (111, 5’-0-acetyl) en route t o I (5’4-acetyl) was varied with the reaction conditions. With glacial plausible. T h e elucidation of t h e structure of the halogeno deoxy nucleoacetic acid as solvent a crystalline monobromodisides (11) in this paper strongly supports their structure for the iodo derivative. mesyloxy nucleoside [VIIIk(Br) ] was obtained. This (15) D. M. Brown, D. B. Parihar, and A. Todd, J. Chem. Soc., 4242 proved to be analogous to the chlorodeoxy nucleoside (1958).
.4 ’
m
(16) D. M. Brown, D. B. Parihar, C. B. Reese, and A. Todd, ibid., 3035 (1958). (17) E. J. Reist, L. Goodman, and B. R. Baker, J . Am. Chem. Soc.. SO, 5775 (1958).
(18) R. Fecher, J. F. Codington, and J. J. Fox, ibid., 88, 1889 (1961). (19) T h e reaction of IXk with sodium benzoate in dirnethylformamide under less strenuous conditions replaces the 5’-mesyloxy group t o give IXj.11
566
CODINQTON, DOERR, AND Box
[VIIIk(Cl)]; it too was readily converted to IXk under conditions using sodium acetate in dimethylformamide. Mesylation of II(Br) in pyridine to give VIIIk(Br) established the structure of II(Br). At elevated temperature with benzene as a solvent, the reaction of IXk with hydrogen bromide yielded a dibromomonomesyloxy nucleoside. Proof that this material was 2‘,5’dibromo-2’,5’-dideoxy-3’-O-mesyluridine [VIIIh(Br)] was obtained by its conversion to IXj in hot acetamidesodium benzoate. Confirmation of the 2‘-position (but not configuration) of the bromo substituent in compound II(Br) was obtained by its reduction to 2‘-deoxyuridine (VII, Scheme I)I6 using palladium on charcoal in absolute ethanolic solution. Isolation of the product proved more difficult than anticipated because 5,6-dihydro derivatives also- were formed in substantial amounts (20-30%). Since the fluorodeoxy nucleosides were formed under considerably more vigorous conditions than either the chloro or bromo derivatives, an independent determination of the structure of II(F) was necessary. Attempts to reduce II(F) to the known 2’-deoxy-5,6dihydrouridine20 failed. The fluoro substituent was not removed after 48 hr. a t atmospheric pressure with palladium on charcoal in ethanol-water. Attempted hydrogenation a t high pressure (more than 100 atm.), with palladium on charcoal in dilute ethanol,21 was likewise unsuccessful. Reaction of the unblocked fluorodeoxyuridine I1(F) with sodium 1-butoxide in dimethylformamideZ2resulted in a 59% yield of crystalline 2,2’-anhydro-l-PD-arabinofuranosyluracil (I, Scheme I). Since I might have arisen from the epoxide 111, an attempt was made to establish the structure of II(F) by the method utilized successfully for II(C1) and II(Br). This ap(20) M. Green and 5. S. Cohen, J. Bid. Chem., 146, 397 (1957). (21) We are indebted t o Dr. Robert Duschinsky of Hoffmann LaRoche, Inc.. Nutley, N. J., for this experiment. (22) R . Letters and A. M. Miohelson, J. Chem. S O L . ,1410 (1961). (23) Mesylation of I 1 (F) gave an amorphous dimeayloxyfluoro derivative [ V I I I k ( F ) I in only fair yield. A n attempt to form this same product from the reaction of I X k a i t h hydrogen fluoride in dioxane at 100-llOo failed. A n amorphous material (B) was obtained. On the baais of i t s infrared and ultraviolet absorption properties i t appeared t o be similar to, if not identical with, V I I I k ( F ) , although subsequent experiments showed t h a t little, if any, material of structure V I I I k ( F ) was present. Attempted crystallization of B from ethanol produced a small amount of colorless crystalline material C with a markedly different spectrum in the infrared from V l I l k ( F ) . Although the general characteristics of the two spectra were related. C exhibited more, and more sharply defined, absorption bands in the 7-10-p region than B which has a n infrared spectrum almost identical with t h a t of trimesyloxyuridine. T h e spectrum of C in the ultraviolet was similar t o t h a t of uridine (max. a t 262 mp, min. a t 230 m p ) ; whereas, B exhibited a hyposchromic shift, a s did VIIIk(F) (max. a t 258 mp, min. a t 229 rnr). Passage of B through a cellulose column with l-butanolwater (86:14) as solvent produced two distinct peaks. The first peak. representing a n approximate yield of 30% baeed upon I X k , contained material (from a small cut a t the center of the peak) with C, H, N , F , and S analyses corresponding to CuHieFzNiOiSz. Although the s t r u c h r e of this material remains unknown, it seems probable t h a t a t least one of the two fluorine atoms is attached t o a sulfur atom of one of the two mesyloxy groups.2‘ Material from the second peak proved t o he identical with C isolated from J3 by crystallization from ethanol. Compound C was found t o be present in the reaction mixture in approximately 6% yield based upon I X k . Elemental analyses ( C , H. N , S, and F) were consistent with a formula of CioHiaF1Nz0sS. I n addition t o the presence of two fluorine atoms, this crystalline product had clearly lost a mesyloxy group. Although the structure of C has not been established, the possibility of a 2’.3’- or 3’,5’-ortho ester type structure (oxygen-sulfur-oxygen bridge) with a fluorine attached tosulfur is suggested. (24) .Ilthough no precedent could be found for a fluorine-substituted methanesulfonyl ester. a review article by H. L. Roberts [Quart. Rev. (London), 16, 30 (1961)l on the chemistry of the sulfur-fluorine bond clearly indicates t h a t such structures would not be improbable.
VOL.
29
proach was unfortunately unsuccessful due to the fact that the reaction of anhydrous hydrogen fluoride with IXk in dioxane gave material not analogous to VIIIk(C1 or Br) but containing more than one fluorine atom.23 Attention was turned to the use of trityl blocking groups. It had been demonstrated that the reaction of trityl chloride with pyrimidine nucleosides in pyridine produces under relatively mild conditions good yields of S’-O-trityl nucleosides. 26 Investigating a report by Levene and Tipsonz6of the isolation of a di-0trityl derivative of uridine, Yung and characterized this material as 2’,5’-di-O-trityluridine. Although these investigatorsz7 were unable to isolate 3’,5‘ditritylated uridine, they did obtain chemical evidence for its presence in small amounts in the reaction mixture. It thus appeared that tritylation of the 3’hydroxyl of nucleosides was feasible. Tritylation of II(F) in pyridine at room temperature gave the 5’-0-trityl derivative (XII, Scheme 111) in crystalline form. As expected, tritylation of the remaining free hydroxyl group proved difficult. After reaction of XI1 with 2 equiv. of trityl chloride for 24 hr. a t 75-looo, 75% of XI1 was recovered unchanged, but paper chromatography with an ethyl acetatemethanol-a-heptane system (solvent A) detected a second compound. After more vigorous reaction conditions, 100-106° for 40 hr., such chromatography revealed the presence of two components, Rr 0.33 and 0.74, in approximately equal amounts. The slower moving component was identified as the mono-0trityl starting material (XII) and the faster component as a di-0-tritylfluoronucleoside (XIII) . Physical properties of the tritylated nucleosides are presented in Table I. Separation of the components of the reaction mixture was achieved with Whatman 3MM paper, solvent SCHEME I11
A
O N
XPI
OH
OH
OH
xpll
XPIII:
P. A. Levene and R. S. Tipson, J. Bioi. Chem., 104,385 (1934). P. A. Levene and R. S. Tipson, ibid.. 106, 419 (1934). (27) N. C. Yung and J. J. Fox, J . A m . Chem. Soc., 8 8 , 3080 (1981).
(25) (26)
MARCH, 1964
2’-HALOGENO-2’-DEOXYPYRIMIDI?JE NUCLEOBIDES
567
TABLE I PHYSICAL CONSTANTS OF TRITYL DERIVATIVES IN ETHANQL [&]Z‘D
XVIII 5’-O-Tritylarabinosyluracil XV 3’,5‘-Di-O-tritylasabinosyluracil XI1 5’-O-Trityl-2’-fluorodeoxyuridine XI11 3’,5’-Di-O-trityl-2’-fluorodeoxyuridine XVII 5’-O-Trityl-2,2’-anhydroarabinosyluracil
+13” f38’
+23” - 12” - 25”
A, and elution of XIII with ether. XI11 was obtained
--
h
261.5 261.5 259 259 248 sh
280/ x (ernax)--
(9800) (10,800) (8970) (9500) (6740)
-Amin
241 244 242 244
(trnid-
(5300) (6550) (5380) (6560)
240
1.81 1.39 1.62 1 20 0.66
described in the previous paper2 are assigned 2’-ribo as a colorless amorphous solid. Proof of the position structures and are 2’-halogeno analogs of thymidine and configuration of the fluoro function required the and 5-fluoro-2’-deoxyuridine. conversion of XI11 to the 2,2’-anhydro derivative (XIV) without loss of the two trityl groups. IsolaExperimental2* tion and structural determination of either XIV or of 2’-Chloro-2’-deoxy-3’,5‘-di-O-mesyluridine[VIIIk(Cl)]. the di-0-tritylarabinosyl nucleoside (XV), which could Method A.-A suspension of 2.55 g. (6.7 mmoles) of 2,2‘-anonly have resulted from hydrolysis of XIV, was neceshydro-1-(3’,5’-di-0-mesyl-,5-~-arabinosyl)uracil (IXk) in 130 ml. sary. Initial attempts to convert XI11 to the 2,2’of dioxane (distilled from potassium hydroxide) saturated with hydrogen chloride was heated in a closed stainless steel container anhydro derivative (XIV) using sodium t-butoxide in a t 70-80” for 19 hr. The amber-colored solution was taken to dimethylformamide2* failed. The high temperature dryness in vacuo, and the gum residue was triturated with 300 (115-118’) necessary for the removal of the fluoro ml. of water, whereupon crystallization occurred. Colorless substituent produced fluorescent products. No chrocrystals, 2.43 g. (87%), m.p. 147-150’, were collected. Crystalmatographic spot attributable to the desired product lization from ethanol gave colorless micaceous plates, m.p. 151-152’, [ a ] ” D +23” (c 0.3, acetone). (XIV) was observed. Anal. Calcd. for CllH16ClN20&2: C, 31.55; H, 3.61; C1, It was hoped that 3’,5‘-di-O-tritylarabinofuranosyl- 8.47; N, 6.68; S, 15.30. Found: C, 31.87; H, 3.56; C1, 8.73; uracil (XV) could be isolated after alkaline hydrolysis N,6.81; S, 15.90. of XIII, a reaction that would necessarily involve the VIIIk(C1). Method B.-To a solution of 0.06 g. (0.23 mmole) of 2’-chloro-2‘-deoxvuridine III(C1)l in 3 ml. of Dvridine at 0” w a ~ formation of the 2,2’-anhydro nucleoside (XIV) added 0.13 g. (1.1 mmoles) of methanesulfonyichloride. After as an intermediate. Proof of the structure of XV remaining a t 0” for 18 hr., the yellow solution was treated with a would establish the structure of II(F). Reflux of few drops of ethanol. Solvent removal zn vacuo left a yellow XI11 with an ethanolic sodium hydroxide solution for gum, which was triturated in turn with 1 : 1 ether-petroleum ether (b.p. 30-60”) and water. The residue w m dried and triturated 15 hr. gave a 40% conversion of XI11 (Rf 0.79, solvent with 2 ml. of ethanol. Pale yellow crystals, 0.09 g., m.p. 145A) to a new di-0-trityl derivative (Rr 0.51). The mix149’, were recrystallized from ethanol to give colorless crystals, ture was chromatographed on Whatman 3MM paper melting a t 150.5-152”, identical (melting point, ultraviolet, and (solvent A) and eluted with ether. Removal of the ether infrared spectra) with a sample prepared by method A. A mixfrom the slower moving component gave a colorless ture of samples prepared by methods A and B showed no depression in melting point. amorphous solid. Elemental analysis and physical prop2’-Bromo-2’-deoxy-3’,5’-di-O-mesyluridine [VIIIk(Br)] , erties (see Table I) were consistent with the ditrityl Method A.-A suspension of 0.95 g. (2.5 mmoles) of IXk in 100 structure (XV). This material was found to be identical ml. of glacial acetic acid was saturated with hydrogen bromide a t (infrared and ultraviolet spectra and chromatographic 20-25’. After remaining in a sealed container a t this temperature, with stirring, for 9 days, the resultant dark red liquid was properties) with the product obtained upon the tritylation of 1-(5’-O-trityl-P-~-arabinofuranosyl)uraci1 taken to dryness in vacuo. The residue was triturated with ether, then with water. Treatment of a hot ethanolic solution with (XVIII). XVIII had been obtained from the alkaline activated charcoal removed all color. Upon cooling, colorless hydrolysis of the 5’-0-trityl-2,2’-anhydronucleoside platelets, 0.54 g. (47%), m.p. 142-146”, were obtained. Crystal(XVII) . 2 lization from 80% ethanol gave micaceous platelets, m.p. 151155”, [ a ] 2 3 D +30° ( c 0.2, acetone). The trityl groups were removed from XV by reAnal. Calcd. for C11H1,Br?J20&32: C, 28.52; H, 3.26; Br, fluxing in hydrochloric acid-water-ethanol. The ul17.26; N, 6.04; S, 13.83. Found: C, 28.59; H, 3.32; Br, traviolet absorptidn spectrum of the residue was that 17.37; X , 6.08; S, 13.93. of a 1-P-D-aldopentofuranosyluracil. Paper electroVIIIk(Br). Method B.-A solution of 0.10 g. (0.33 mmole) of 2’-bromo-2’-deoxyuridine [II(Br)] in 3 ml. of pyridine was phoresis in borate buffer of pH 9.3 revealed only one cooled to 0-5”. After the addition of 0.14 g. (1.3 mmoles) of spot, which migrated with an authentic sample of 1methanesulfonyl chloride, the mixture was allowed to remain in P-D-arabinofuranosyluracil (XVI) but a t a different the cold for 18 hr. The product was isolated in essentially the rate from the xylo, lyxo, or ribo isomers. same manner as described above for VIIIk(Cl), method B. Colorless needles, 0.03 g., m.p. 141-150°, were obtained by crysSince XVI could have been formed from XI11 only tallization from ethanol. An infrared spectrum (potassium brothrough the 2,2’-anhydro intermediate (XIV) the isomide pellet) of this material was identical in every respect with a lation of XVI as the only 1-P-D-aldopentofuranosyl- sample prepared by method A. uracil present proves conclusively both the position and 2 ’,5 ’-Dibromo-2 ’,5 ’-dideoxy-3 ’-0-mesyluridine [VIIIh(Br ) ] .A mixture of 1.0 g. (2.6 mmoles) of IXk in 100 ml. of dry benzene the configuration of the fluoro substituent in II(F), and firmly establishes I1(F) as 2’-fluoro-2’-deoxyuri(28) Melting points are corrected. Elemental analyses a e r e made by the Spang Microanalytical Laboratory, Ann Arbor, Mich. Ultradine. violet absorption d a t a were obtained on the Cary recording spectrophoBased upon similar chemical, physical,2 and biotometer, Model 15, and the Beckman spectrophotometer, Model DU. Infrared determinations were made with t h e Infracord spectrophotometer logical13 properties, and upon the similar reaction by use of the potassium bromide disk technique. Paper electrophoretic conditions under which they are formed,* the other separations were made with a n apparatus manufactured by the E. C. Aphalogeno (chloro, bromo, and fluoro) deoxy nucleosides paratus Co.
568
CODINGTON, DOERR, AND Fox
was saturated with hydrogen bromide a t 0-5'. The reaction mixture waa heated in a sealed stainless steel container at 75-80" for 17 hr. After removal of solvent in vucuo, the dark-colored residue waa triturated in turn with ether and water. A hot ethanol-water (4:1) solution was treated with activated charcoal and filtered. Colorless needles, 0.61 g. (51%), m.p. 197-205', separated. Crystallization from 90% ethanol gave colorless D (c0.5,acetone). needles, m.p. 198-204" dec., [ c Y ] ~ ~+IS0 Anal. Calcd. for C10H12BrZNz06S: C, 26.80; H, 2.69; Br, 35.67; N , 6.25; S, 7.15. Found: C, 26.75; H , 2.87; Br, 35.38; N, 6.10; S, 7.26. 2.2'-Anhvdro-1-~4-~-arabinofuranosvl~uraci1 (1) from II(F).A mixture bf 25 mg. (0.10 mmole) of'II(F), 0.'30 ml. of tibutyl alcohol containing 0.15 mmole of sodium t-butoxide and 2.5 ml. of dimethylformamide waa heated a t 101)-110' for 2 hr. The cooled mixture was triturated with 50 ml. of ether-petroleum ether (2:3). The residue was taken up into hot ethanol (95%); the solution waa decolorized with activated charcoal and cooled. Colorless crystals, 11 mg. (59%), m.p. 244-246", were collected. This material waa identical (melting point, ultraviolet, and infrared spectra) with an authentic sample. Admixture with an authentic specimen showed no depression of melting point. 2,2'-Anhydro-1-(5 '-0-benzoyl-3 '-0-mesyl-p-D-arabinosy1)uracil (IXj). Method A.-A mixture of molten acetamide (1.8 g.), 0.1Og. (0.7 mmole) of sodium benzoate, and0.10 g. (0.20 mmole) of VIIIj(Br) was heated a t 100-105' for 20 rnin., then poured into 50 ml. of ice-water. After 30 rnin., crystals were collected, washed with water, then ether. The nearly colorless needles, 0.052 g. (64%), melted at 236-238'. Admixture with an authentic sample showed no depression and the two samples were identical with respect to ultraviolet and infrared spectra. IXj. Method B.-A mixture of 19 mg. (0.043 mmole) of VIIIj(C1) and 34 mg. (0.23 mmole) of sodium benzoate in 0.30 g. of acetamide was heated at 100" for 45 min. To the colorless solution waa added 5 ml. of water. After 30 min. colorless crystals, 11.0 mg. (63%), m.p. 229-235', were collected. Crystallization from SOYo ethanol gave colorless needles melting a t 237238". Mixture of this material and an authentic specimen of VIIIj showed no depression of melting point. IXj. Method C.-To 0.20 g . (1.4 mmoles) of sodium benzoate in 2.0 g. of molten acetamide wm added 0.10 g. (0.22 mmole) of 2'5'-dibromo-2',5'-dideoxy-3'-O-mesyluridine [VIIIh(Br )] . The mixture waa heated a t 120-125" for 20 rnin., then poured into 30 ml. of icewater. Tan crystals were collected and crystallized from ethanol-water (1: 1). Colorless needles, 0.02 g. (227,), m.p. 233-235", identical (infrared and ultraviolet absorption spectra) with an authentic sample of IXj, were obtained. 2,2 '-Anhydro-1-( 3 ' 3 '-di-O-mesyl-P-D-arabinosyl)uracil( I n ) . Method A.-A mixture of 0.10 g. (1.2 mmoles) of anhydrous sodium acetate and 0.10 g. (0.24 mmole) of VIIIk(C1) in 6 ml. of dimethylformamide waa heated at 110-115" for 30 min. The solvent was removed in vucuo, the gum residue waa triturated with ether, and the ether was decanted. The material was dissolved in 0.5 ml. of hot ethanol-water ( 1 : l ) and cooled. The colorless needles, 0.006 g. (7%), m.p. 180-182', collected were identical (infrared and ultraviolet absorption spectra) with an authentic sample of VIIIk. IXk. Method B.-To a suspension of 0.20 g. (2.4 mmoles) of anhydrous sodium acetate in 15 ml. of dimethylformamide was added 0.20 g. (0.43 mmole) of VIIIk(Br). The mixture was heated at 100-110" for 40 min. Essentially the same procedure described for the isolation of IXk in method A (above), produced colorless needles, 0.006 g. (4%), m.p. 192-194°.29 These were identical (infrared and ultraviolet spectra) with an authentic sample. A mixture of this material and an authentic sample gave no melting point depression. 2'-Fluoro-5'-0-trityl-2'-deoxyuridine (XII).-To a solution of 0.60 g. (2.4 mmoles) of II(F) in 12 ml. of pyridine, 1.36 g. (4.9 mmoles) of trityl chloride was added with stirring. The solution was stirred at 20-25" for 16 hr., then warmed at 60" for about 15 min. Ethanol (0.3 ml.) was added. After 30 rnin., removal of solvent in vacuo left a yellow gum that was triturated well with water. The water then was decanted. After drying, the material was triturated in ether-petroleum ether and filtered. The colorless crystals, 0.45 g. (387,), m.p. 186-194", obtained were
(29) Sinoe t h e melting point of IXk varies with the r a t e of heating," the difference in melting point ( 1 2 O ) found for IXk,prepared by t h e two different methods (A and B). is not surprimng.
VOL. 29
recrystallized from ethanol to give colorless platelets melting at 199-201". Polarimetric and spectral data are found in Table I. Anal. Calcd. for C2aH26FNz06: C, 68.84; H, 5.16; F, 3.88; N, 5.74. Found: C, 68.46; H, 5.39; F, 3.54; N, 5.10. 2 '-Fluoro-3',5 '-di-O-trityl-Z '-deoxyuridine (XIII).-A solution of 0.26 g. (0.53 mmole) of XI1 and 0.30 g. (1.1 mmoles) of trityl chloride in 3 ml. of pyridine waa heated for 2 hr. at 78", 31 hr. at 89-95", and 7 hr. a t 105-106'. To the dark amber-colored solution waa added 0.2 ml. of ethanol. After 10 min. the still warm solution waa poured with stirring into 200 ml. of ice-water. The tritylated nucleosides were extracted with chloroform, and the chloroform solution was extracted in turn with water and dried. The solvent, removed in vucuo, left a residue of 0.24 g. Paper chromatography with a system of n-heptane-ethyl acetatemethanol, solvent A,Z revealed two spots of about equal intensity under ultraviolet light representing the mono-0-trityl ( R f0.55) and the di-0-trityl (Rf 0.83) derivatives. The entire product waa separated in this fashion (Whatman 3MM paper, solvent A). The di-0-trityl nucleoside ( X I I I ) waa eluted with ether, and the ether was removed in vacuo. The residue was triturated with water, filtered, and dried. The yield of colorless amorphous solid was 0.10 g. (26%). Polarimetric and spectral data are presented in Table I . Anal. Calcd. for C4,H8sFN2O6:C, 77.24; H, 5.38; F, 2-60; N,3.82. Found: C, 77.18; H, 5.50; F , 2.95; N , 4.29. 1-( 5 '-0-Trityl-p-D-arabinofuranosy1)uracil(XVIII).-The solution resulting from the addition of 3 ml. of 1 N sodium hydroxide to a suspension of 0.80 g. (1.7 mmoles) of 2,2'-anhydro-1-(5'-0trityl-p-D-arabinofuranosyl)uracil(XVII) in 40 ml. of ethanolwater (1 : 1)was stirred for 1 hr. a t 20-25', then neutralized by the dropwise addition of 2 N acetic acid. The product was collected, triturated with water, and crystallized from ethanol-water (1: 1) to produce 0.72 g. (87%) of a colorlesssolid melting with effervescence a t 119-120°. Spectral and polarimetric data are found in Table I. Anal. Calcd. for Cz8Hz6X206: C, 69.12; H, 5.39; N, 5.76. Found: C, 68.00; H, 5.43; N , 5.79. 1-(3 ' 3 '-Di-O-trityl-p-~-arabinofuranosyl)uracil (XV) . Method A,-To a solution of 10 mg. of XI11 in 7 ml. of 70% ethanol waa added 1 ml. of N sodium hydroxide. The mixture waa refluxed for 15 hr. After the removal of the solvent in vacuo, the residue was suspended in 8 ml. of water, and the solution waa made slightly acidic by the addition of 1 drop of acetic acid. The solid waa collected, washed well with water, and dried. Paper chromatography (ascending, S and S No. 597) with solvent A gave two spots, Rt 0.51 and 0.79, representing the di-0-tritylarabinosyl derivative (XV) and starting material (XIII), respectively. The entire product waa chromatographed on Whatman paper (3MM) to give a clear-cut separation. The XV-containing strip waa eluted with ether. After removal of solvent an uacuo, the residue was triturated with water, filtered, and dried. This sample was identical with the material prepared by method B (see below) with respect to ultraviolet and infrared absorption properties and chromatographic behavior in solvent A. XV. Method B.-A solution of 0.08 g. (0.16 mmole) of XVIII and 0.11 g. (0.39 mmole) of trityl chloride in 2 ml. of pyridine was heated at 90" for 15.5 hr. A few drops of ethanol were added to the mixture. Solvent waa removed in vacuo, and the residue waa triturated with water. After filtration, the solid waa dried and triturated well with two 20-ml. portions of hot n-heptane. The solid waa collected and dried. The colorless amorphous material weighed 0.08 g. (62%). Chromatography (Whatman 3MM paper, solvent A, ascending) of 0.06 g. gave a complete separation of the material into three components characterized aa tritylethyl ether (migrating near the front), di-0-tritylspongouridine (XV, Ri 0.6-0.8), and mono-0-tritylspongouridine (XVIII, R f0.2-0.5). The middle strip containing XV was eluted with ether, and the ether waa removed in vacuo. After trituration with water, the colorless amorphous residue waa collected and dried. The product, 0.016 g., m.p. 145-160", proved identical with a sample obtained by method A. Spectral and polarimetric data are found in Table I. Anal. Calcd. for Ca,HloNzOe: C, 77.45; H, 5 53; N, 3.94. Found: C , 76.60; H , 5.81; N, 3.92. Detritylation of XV.-A solution of approximately 2 mg. of XV, prepared by method A, and 0.50 ml. of 1 N hydrochloric acid in 2 ml. of ethanol was refluxed for 18 min. The resulting solution was taken to dryness and distilled five times with portions of benzene in vacuo. The residue waa triturated with a little ether,
MARCH,1964
SYNTHESIS OF PYRIDINES
and the ether WM decanted. This residue WEB dissolved in 2 drops of water. The ultraviolet absorption spectrum in water (max. a t 262.5 mp,min. at 230 mH) corresponded to that of a 1-8o-aldopentofuranosyluracil. Electrophoresis ( Whatman 3MM paper, borate buffer, 0.07 M , PH 9.2, 800 v., 2.5 hr.1 revealed only one spot of high intensity under ultraviolet light. Its anodic migration was 3.5 cm., which corresponded to that of a test sample of l-p-D-arabinofuranosyluracil (XVI) of anodic
569
migration, 3.5 cm. No absorption waz found which corresponded to that of a test sample of l-8-D-XylOfUranOSylUraCil (anodic migration 13.5cm.).
Ac~owledgment.-~he authors wish to express their appreciation to Dr. George B*Brown for his suggestions and continued interests.
Synthesis of Pyridines from Butadiene and Cyanogen-Like Molecules GEORQE J. JANZ AND ALANR. M O N A H A N ' ~ Department of Chemistry, Rensselaer Polytechnic Institute, Troy, New York Received September 86, 1963 Experiments are described using a series of "cyanogen-like'' molecules, Le., CFsCFzCN, CF3CF2CF2CN, CFzCICN, and CFCl?CN, as the dienophiles for the synthesis of new pyridines from 1,3-butadiene. The gas phase reaction proceeds smoothly a t moderately high temperatures (350-450") and atmospheric pressures. Pyridinic products are formed in very nearly 100% yields based on the nitrile having fluorine as the only substituent.
The formation of 2-perfluoromethylpyridine through the addition of perfluoroacetonitrile and butadiene and the reaction kinetics of this cyclization reaction in the homogeneous gas phase have been reported . l b , O As with cyanogen, the addition reaction proceedsId a t moderately high temperatures (35CL400") a t atmospheric pressures in the homogeneous gas phase. The present work reports the results for experiments with related "cyanogen-like" molecules as dienophiles [e.g., CF3CF2C1;, CF3(CF&CN, CF2C1C?;, and CFC12CN] in this new pyridinic synthesis. Experimental Chemicals.-Butadiene, 99.0% purity, b.p. -3.0" (Matheson Co., C.P. grade), was degassed three times (-195') prior touse. The nitriles, CFICN, b.p. -60"; CFICFZCN,b.p. -35"; CF3(CF2)&N, b.p. -5'; CFzCICN, b.p. -18.3'; and CChFCN, b.p. 35', each obtained in 98% purity (Peninsular ChemResearch Inc., Columbia Organic Chemicals Co.), were similarly treated. Apparatus and Procedure.-The continuous flow assembly, previously described,ld was modified to take a large reaction vessel (5 1.). The latter was pretreated with an 8-hr. oxygen "sweep" (100 cc./min.) a t 500', prior to each experiment. The whole assembly was flushed with an inert gas (dry nitrogen, 2 hr., 100 cc./min.) after the oxygen pretreatment and prior to an experiment; on completion of each, run a similar nitrogen "chase" was used to recover the materials remaining in the system. Gaseous reactants were monitored by precision Manostat flowmeters (Model No. F M 1043 B). A continuous feed injection assembly was used for liquid reactants (e.g., CCLFCN). This consisted of a 50-cc. Perfektum syringe, gear driven a t a constant rate (0.16 mole/hr.) through a Borg 3-phase motor (30 r.p.m.) and micrometer screw. Table I summarizes the experimental conditions, material balances, conversions of reactants per pass, and yields of pyridinic products. Deviation of the over-all material balance (Table I) from 100% is attributed to experimental losses due to transfer steps in the crude product analysis (weighings, low temperature fractionations). The composition of the distillate was quantitatively established by vapor phase chromatography using 550silicone and di-n-decyl phthalate columns. The refractive indices and boiling points for the three new pyridinic products together with the values for 2-perfluoroalkylpyridine are as follows: (1) 2-pentafluoroethylpyridine, 72% 1.3949, b.p. 151'; (2) 2heptafluoro-n-propylpyridine, naso 1.3814, b.p. 162'; (3) 2difluoromonochloromethylpyridine, nZ3o 1.4646, b.p. 162'. ( l ) ( a ) Union Carbide Fellow in Chemistry, 1962-1963; (b) G. J. Janz, J. M. S. Jarvie, and W. E. Fitzgerald, J . Am. Chem. SOC.78, 978 (1956); (c) G. J. Jana and J. M. S. .Jarvie, J . P h y s . Chem. 60, 1430 (1956); (d) G . J. J a n s and M. A. DeCrescente, J . Ore. Chem. 18, 765 (1958).
TABLE I SYNTHESIS O F 2-PERHALOALKYLPYRIDINES Run conditione
CFICN
Contact time, hr. 0.44 Total time, min. 125 Temperature, 'C.(*3") 400 Butadiene, moles 0.19 RCN, moles 0.24 Material Recovery, wt. % 98
CFaCFzCN
0.46
CFiCFaCFlCN
0.49
CFnClCN
0.38
107
120
115
400
400
400
0.17 0.18
92
0.19 0 17
92
0.31 0.19
100
Conversions per single pass, mole %
Butadiene RCN
71 45
54 33
89 22
59 31
Yields of pyridinic product, mole %
A" Bb
79 62 22 99 97 97 a Yield of pyridinic based on converted diene. pyridinic based on converted nitrile.
4 12 Yield of
The physical constants for 2-methylpyridine ( n Z 51.4155, ~ b.p. 143') agreed with those previously reported .lb Both 3-vinylcyclohexene-1 and styrene were confirmed as byproducts but were not quantitatively assessed. An azeotrope (b.p. -124') containing 0.32 mole % of 2-trifluoromethylpyridine and 0.68 mole 90of 3-vinylcyclohexene-1 was found to be present in the crude resulting from the reaction between CF&F and butadiene at 400'. N o azeotropes were detected in the experiments with the other nitriles. The experiments with CFC12CN and 1,3-butadiene were less successful than for CFzClCN (Table I). The pyrolysis and side reactions were very extensive. While separation of the pyridinic product in trace amounts appeared possible, these experiments were not taken any further.
Results Confirmation of the pyridinic products was through infrared and mass spectra, n.m.r. analysis, and microelemental combustion analysis for nitrogen. The results were as follows. Infrared Spectra.-All spectra were gained with a Perkin-Elmer Model 21 double beam recording spectrophotometer having sodium chloride optics. A cell thickness of 0.025 mm. was used.
